Roanoke Times
Copyright (c) 1995, Landmark Communications, Inc.
DATE: SATURDAY, December 10, 1994 TAG: 9412130012
SECTION: VIRGINIA PAGE: B4 EDITION: METRO
SOURCE: ASSOCIATED PRESS
DATELINE: LYNCHBURG LENGTH: Medium
Mild-mannered Thomas Nicely was content with his quiet life as a mathematics professor at Lynchburg College, toiling 60 hours a week teaching freshmen, grading exams and researching number theory.
Then, he found a flaw in Intel Corp.'s top-of-the-line Pentium microprocessor and publicly embarrassed the world's largest computer-chip maker. And now, he's enjoying - or at least tolerating - his moment of fame.
``Mathematicians in general have very private lives,'' Nicely drawled in a thick West Virginia accent. He was a coal miner's son and the first in his family to go to college.
Nicely discovered through an equation on his math department's three-Pentium system that Intel's chip flubbed some calculations. It took him months to prove the chip was faulty, and the finding quickly became public. The Santa Clara, Calif.-based company, which had actually discovered the error on its own but decided not to go public with it, has apologized.
``He obviously knows his mathematics,'' Intel spokesman John Thompson conceded.
Now, when Nicely's phone rings, it's The Wall Street Journal, The New York Times, CNN or National Public Radio on the line.
The story of this unlikely celebrity began in June, when he noticed that on some division problems, his computers churned out faulty numbers beginning with the ninth place to the right of the decimal point. That may seem very minor, but ``to me, it was a huge error,'' Nicely said.
He checked and rechecked his own equation but couldn't find anything wrong with it.
But Lynchburg College, a private, 1,200-student liberal arts school with no graduate program in math, isn't exactly M.I.T. or Stanford as far as computer facilities go. Nicely used computers at an office supply store to test his hypothesis that the problem was in the chip.
When the error surfaced on the store's computers, he knew he was right.
Magazines from the Netherlands, France and Great Britain and an Australian newspaper also have called.
On Oct. 24, Nicely called Intel's technical support line and was informed no one knew of the problem. He left messages and called back, but six days went by and he heard nothing.
That's when Nicely e-mailed several colleagues, presenting them with his findings. One of the people he messaged decided to post the findings to the world on the Internet, a global telecommunications network.
``Then suddenly I heard from Intel again,'' Nicely said, laughing. ``They seemed very anxious to please me at that point.''
Intel officials then asked him to send them a copy of his program. The company also sent him two new chips, which he says have been running his research program flawlessly.
Nicely, who attended West Virginia University as an undergraduate and got his doctorate from the University of Virginia, said he's surprised the bug wasn't discovered by someone else, especially a researcher at a major institution.
``It indicates to me that a lot of people are not checking their work.''
In fact, Intel's own technicians detected the error in July. But the company decided not to go public with the information, reasoning that only a slight number of users would ever encounter the problem.
Nicely won't accept that explanation.
``It's a very simple-looking problem that it's getting wrong,'' Nicely said.
Nicely points out that many computer chips have had bugs, but what sets this case apart is the company's response to the problem. Intel's replacement offer extends only to users directly affected by the flaw. A handful of transistors added to the new chips solves the problem, Intel said.
Nicely and Intel may be making amends, however. They have discussed the possibility of Nicely's working as a free-lance consultant. Nicely hopes Lynchburg College would get free computers out of the deal.
by CNB